Resolution bound and detection results for scattering centers

This paper is concerned with detection and estimation of the scattering centers of a target from coherent, stepped frequency measurements. It presents the asymptotic (as SNR→∞) probability density function for the exponential model parameter estimates using the total-least-squares Prony algorithm. This asymptotic function is used to derive scattering center resolution and detection bounds for the TLS-Prony algorithm, for comparison with Cramer-Rao bound results. Monte-Carlo simulations are also compared with the theory. The resolution bounds are obtained from the standard deviation bounds of the pole angles in the exponential models. The detection bounds are obtained by considering the probability that the energy of an estimated mode exceeds a pre-defined threshold. It is shown that for sufficiently high SNR, both the CRB and TLS-Prony resolutions are better than can be obtained using the FFT